From af4d2042e1bf86515a47b109e967cab7355b297a Mon Sep 17 00:00:00 2001
From: Dengddd <3149942468@qq.com>
Date: Tue, 8 Oct 2024 21:42:36 +0800
Subject: [PATCH] =?UTF-8?q?=E6=9B=B4=E6=96=B0=20function.py?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 function.py | 119 +++++++++++++++++-----------------------------------
 1 file changed, 38 insertions(+), 81 deletions(-)

diff --git a/function.py b/function.py
index d475eb2..fe0cfc3 100644
--- a/function.py
+++ b/function.py
@@ -38,84 +38,41 @@ name = "邓龙树"
 - int: 预测的分类标签。
 """
 
-import numpy as np  
-  
-def mahalanobis_distance(point1, point2, cov_inv):  
-    """计算两个点之间的马氏距离。  
-      
-    Args:  
-    - point1 (numpy array): 第一个点。  
-    - point2 (numpy array): 第二个点。  
-    - cov_inv (numpy array): 协方差矩阵的逆。  
-      
-    Returns:  
-    - float: 马氏距离。  
-    """  
-    diff = np.array(point1) - np.array(point2)  
-    return np.sqrt(np.dot(np.dot(diff, cov_inv), diff.T))  
-  
-def compute_mahalanobis_distances(dataSet, inX, cov_inv):  
-    """计算数据集中每个点与输入点inX的马氏距离。  
-      
-    Args:  
-    - dataSet (numpy array): 训练数据集。  
-    - inX (numpy array): 输入待分类的样本。  
-    - cov_inv (numpy array): 协方差矩阵的逆。  
-      
-    Returns:  
-    - numpy array: 马氏距离数组。  
-    """  
-    inX_array = np.array(inX)  
-    distances = [mahalanobis_distance(dataSet[i], inX_array, cov_inv) for i in range(len(dataSet))]  
-    return np.array(distances)  
-  
-def classify0(inX, dataSet, labels, k=3, cov_inv=None):  
-    """KNN 分类算法，使用马氏距离。  
-      
-    Args:  
-    - inX (list or numpy array): 输入待分类的样本。  
-    - dataSet (numpy array): 训练数据集。  
-    - labels (numpy array): 训练集对应的标签。  
-    - k (int): KNN 算法参数，表示选择最近的k个邻居，默认值为3。  
-    - cov_inv (numpy array): 协方差矩阵的逆，如果为None则使用欧氏距离。  
-      
-    Returns:  
-    - int: 预测的分类标签。  
-    """  
-    if cov_inv is not None:  
-        distances = compute_mahalanobis_distances(dataSet, inX, cov_inv)  
-    else:  
-        distances = compute_distances(dataSet, inX)  # 使用原有的欧氏距离计算函数  
-      
-    sorted_indices = np.argsort(distances)  
-    k_nearest_indices = sorted_indices[:k]  
-    k_nearest_labels = labels[k_nearest_indices]  
-    unique_labels, counts = np.unique(k_nearest_labels, return_counts=True)  
-    most_common_label = unique_labels[np.argmax(counts)]  
-    return int(most_common_label)  
-  
-# ... (load_data 和 test_knn 函数保持不变)  
-  
-if __name__ == "__main__":  
-    train_file_path = "path/to/datingTrainSet2.tsv"  # 替换为您的训练文件路径  
-    test_file_path = "path/to/datingTestSet2.tsv"    # 替换为您的测试文件路径  
-      
-    # 加载数据  
-    train_data, train_labels = load_data(train_file_path)  
-    if train_data is None or train_labels is None:  
-        print("Failed to load training data.")  
-        exit()  
-      
-    # 计算协方差矩阵及其逆  
-    cov_matrix = np.cov(train_data.T)  
-    cov_inv = np.linalg.inv(cov_matrix)  
-      
-    # 测试KNN算法，使用马氏距离，默认k=3  
-    accuracy = test_knn(train_file_path, test_file_path, k=3, cov_inv=cov_inv)  
-      
-    # 注意：由于test_knn函数中没有直接传递cov_inv，我们需要修改它以便能够传递这个参数。  
-    # 下面的修改是假设在test_knn函数中添加了一个cov_inv参数，并在调用classify0时传递了它。  
-    # （实际代码中需要对test_knn函数进行相应修改，这里只是说明思路）  
-    # ...  
-    # 修改后的test_knn函数调用示例（需要在函数定义中添加cov_inv参数）：  
-    # accuracy = test_knn(train_file_path, test_file_path, k=3, cov_inv=cov_inv)
\ No newline at end of file
+import numpy as np
+
+def mahalanobis_distance(point1, point2, inv_covmat):
+    """计算两个点之间的马氏距离。"""
+    diff = np.array(point1) - np.array(point2)
+    return np.sqrt(np.dot(np.dot(diff, inv_covmat), diff.T))
+
+def compute_mahalanobis_distances(dataSet, inX, inv_covmat):
+    """计算数据集中每个点与输入点inX的马氏距离。"""
+    distances = np.array([mahalanobis_distance(inX, dataSet[i], inv_covmat) for i in range(dataSet.shape[0])])
+    return distances
+
+def classify0(inX, dataSet, labels, k=3):
+    """KNN 分类算法，使用马氏距离。"""
+    # 计算协方差矩阵的逆矩阵
+    covMat = np.cov(dataSet.T)
+    inv_covMat = np.linalg.inv(covMat)
+    
+    distances = compute_mahalanobis_distances(dataSet, inX, inv_covMat)
+    sorted_indices = np.argsort(distances)
+    k_nearest_indices = sorted_indices[:k]
+    k_nearest_labels = labels[k_nearest_indices]
+    unique_labels, counts = np.unique(k_nearest_labels, return_counts=True)
+    most_common_label = unique_labels[np.argmax(counts)]
+    return int(most_common_label)
+
+# 其余函数不变，包括加载数据的 load_data 和测试 KNN 算法的 test_knn
+
+if __name__ == "__main__":
+    train_file_path = "path/to/datingTrainSet2.tsv"  # 替换为您的训练文件路径
+    test_file_path = "path/to/datingTestSet2.tsv"    # 替换为您的测试文件路径
+
+    # 测试KNN算法，默认k=3
+    test_knn(train_file_path, test_file_path)
+
+    # 您可以尝试不同的k值来查看对准确率的影响
+    # test_knn(train_file_path, test_file_path, k=5)
+    # test_knn(train_file_path, test_file_path, k=1)